You are browsing the archive for 2014 March.

OSM data updates

March 11, 2014 in ongoing work

As we have now launched our map tiles at the scale of the Barrington Atlas (the majority of which is at 1: 1 million and 1: 5 million scale), we are now moving on to update and refine our offerings with data from OpenStreetMap (OSM) along with satellite data from OSM and Google.


OSM coastline in Red; Current coastline in black


OSM vs. Current coastlines and rivers

We start by identifying features where OSM and satellite imagery provide a smaller scale and more precise representation of geography than what we currently offer, as seen in the screenshots on the right. We then consult the Barrington Atlas and other information sources to see if there are any significant changes from the ancient world to the present, and then adjust the data accordingly to conform both to visible geography and to the work of the center. This includes removing docks, modern terrain changes, dams, river course changes, etc. In addition to more precise representations of actual geography, such work will also allow us to assign OSM IDs to our features, presenting further opportunities to exploit linked data principles in our data set. These feature changes will also be added to the AWMC API, which provides stable URIs for each feature. As with all of our offerings, Pleiades IDs will also be linked to each feature where appropriate, which will eventually filter into the Pelagios network.

We are beginning the process in Crete, and we will push the results to the public tiles as soon as they are finished. As this is a painstaking process, it will be some time before we cover the entire Mediterranean world. One difficulty is the adjustment of both the coastline and water tiles from their raw OSM forms – we may offer the openwater overlays first, and then update our coastlines once that phase of the project is completed. This will also allow us to roughly define seas, bays, and other water features and therefore provide more complex geometry beyond a representative centroid for such features.

As usual, we welcome comments and suggestions on this work.